05:48 tiredchiku[d]: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12523
05:48 tiredchiku[d]: will attach a backtrace once I figure out how to get it e-e
12:15 phomes_[d]: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32005 landed yesterday and is causing problems on a bunch of games on nvk
14:07 phomes_[d]: looks like it could be in NAK: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12524
16:34 gfxstrand[d]: Yeah, we go unstructured. Maybe we don't unset the bit?
16:36 gfxstrand[d]: See if unsetting that bit right around here fixes it:
16:36 gfxstrand[d]: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/src/nouveau/compiler/nak_nir_lower_cf.c?ref_type=heads#L476
16:36 gfxstrand[d]: phomes_[d]: ^^
16:53 phomes_[d]: gfxstrand[d]: It was already unset on line 455. I printed new_impl->structured when we return from lower_cf_func and it is still false there
16:56 mhenning[d]: I think I have a fix for it
17:12 mhenning[d]: I lied, that fix didn't work right
17:12 tiredchiku[d]: the duality of programming
17:13 karolherbst[d]: what about "I've recompiled and now it does fix it"
17:58 phomes_[d]: mhenning[d]: it did work enough that the games that were broken before are no longer crashing
18:12 mhenning[d]: Yeah, the latest version of https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33219 has passed the first ~45 mins of cts at this point, so I'm cautiously hopeful that fixes all of the issues
18:13 mhenning[d]: Waiting for the full cts run before I unmark draft though
20:09 mhenning[d]: Okay, https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33219 passes cts and should fix the regressions, if anyone is in the mood for a quick review
21:53 gfxstrand[d]: Left you one comment. Otherwise looks okay.
21:53 gfxstrand[d]: mhenning[d]: Left you one comment. Otherwise looks okay.
21:57 phomes_[d]: I have tested it on a few games. Works well in those